Abstract: Compared with traditional robots, biomimetic robots, by drawing inspiration from the morphological structure and behavioral mechanisms of natural organisms, exhibit advantages in adaptability, interactivity, and energy efficiency. In recent years, with the development of embodied intelligence, humanoid, canine-like, and other robots have made significant progress in core capabilities such as perception, decision-making, and continuous learning by leveraging methods including multimodal large models, imitation learning, and world models. Therefore, promoting the application of embodied intelligence across diversified biomimetic platforms will become a key trend in the future. Firstly, we focus on biomimetic hardware and provides an overview of the key technologies of current biomimetic robots in terms of morphological structure, sensing hardware, computational mechanisms, and actuation. Secondly, we outline the technological path from biomimetic morphology to biomimetic intelligence. Finally, we summarize the core challenges faced by current research and outline the possible future research directions, aiming to provide a reference and inspiration for theoretical exploration and technical practice of intelligent biomimetic robots.